Transcript
josh MacDonald: I'm Josh McDonald.
Miranda Materi: I'm Miranda Materia and we are Hand Therapy Academy.
josh MacDonald: Today we're talking about AI and AI in the therapy world. It takes a couple of different forms. So we'll talk about a couple different aspects of it. Let's start with when patients come in and say they've asked ChatGPT or Claude or whatever about their plan of care. And they say, this is what AI tells me. This is what I'm supposed to do.
josh MacDonald: What's your response to that? What do you think, Miranda?
Miranda Materi: um well this happens actually pretty frequently and i think it's a good opportunity to provide education and sometimes they're right you know sometimes you're like okay yeah they'll be like oh i put it in i wanted to know what my next progression was and then i say yeah this is we are going to progress you with the strengthening however we don't want to do you know the strengthening where your cmc joint is subluxing you know i feel like they still need those verbal and physical cues on how to do it properly so in terms of job security i think we're fine but i I don't mind that they look stuff up. but I like that they're... informed and they want to know more so i think it lends itself well to an educational moment.
Miranda Materi: How about you?
josh MacDonald: Yeah, um I try not to get triggered. Yeah. I find um ah more often than not, the patient who goes and looks at things on AI, has thoroughly bought into whatever AI tells them it should be done.
josh MacDonald: And it's it's ah it's a blend between the same thing would happen before AI. And it was my uncle is a doctor or my aunt is a nurse or so -and -so told me this. And so this is what I've been doing.
josh MacDonald: And so it's this education moment where I say, okay, here's the things that we're going to keep from that. And here's the things that are valuable. but Just as a heads up, we're not going to start with a stress ball and strength thing right off the bat. Like we're going to. Slow roll those things. Or it just doesn't have the space for all the nuance of what we're going to do. So, yes, that's the progression.
josh MacDonald: But we're going to talk about waiting until your tissues are ready. And AI just doesn't know that because it's not here for it. so I am. I... treat it as like an educational moment to like, Back.
josh MacDonald: Their pace, usually it's about them going faster and slowing them down a little bit.
Miranda Materi: Yeah, I will say that's true. I think it always lends to them doing more than they probably should early on. But I think that's, you know like I said, a good opportunity to tell them. And good for us too, right?
josh MacDonald: Yeah, yeah
Miranda Materi: Because it's always good to be challenged.
josh MacDonald: Yes, yeah.
Miranda Materi: yeah
josh MacDonald: and And I encourage patients, it I make it well early about their medical care and where they're going with things and all of that. But I also, if it's the patient who can be receptive about a little like joking back and forth banter, and we joke about it a little bit, I would say if...
josh MacDonald: I wouldn't trust my medical care to someone to to a device that could get it wrong. and you didn't you wouldn't know. right like How do you know if there's two or three errors in that? The same thing, I'm not going to trust my 401k decisions to chat GPT. Maybe it's wrong. I don't know. Do I know if it's right or not? And if I know, then I probably don't need to ask. So just kind of like. Helping usually it's...
josh MacDonald: In the clinic I'm in, it's usually our significantly older patients and they are very prone to like. scams and, um you know, being predators coming in for them for their money and stuff.
Miranda Materi: Yeah.
josh MacDonald: So it's like just kind of. Pump the brakes a little bit on buying into any one thing so dramatically. um just because it's you don't know if it's wrong or not. And so you could check with someone who knows. Yeah.
Miranda Materi: Yeah. Yeah, I had one patient. She was actually... a researcher. at the university and she would always do use it was chad gpd and then she'd go into deep research mode which i didn't even know was a thing but apparently it's where you can go deep and it takes like you know an hour or two to get the response or whatever but she would come in with some very interesting like articles and things like that um but there would be mistakes right so i think
josh MacDonald: Okay.
josh MacDonald: Okay.
Miranda Materi: We've both seen that where there's mistakes where it's like, oh, yeah, no, actually, that's not correct or that's not accurate. So I think. It's also a moment to tell people about there's other sources that are more evidence -based than say like ChatJPT or CLAUD.
Miranda Materi: There's some that are more based for medicine and medical care too. so
josh MacDonald: Yeah. Okay, so that's a good segue into our next section on AI is how do we feel about therapists or... other healthcare providers using it for finding out what to do with the patient, using it for protocols or plan of care for patients.
Miranda Materi: yeah I think if you're using an evidence -based one. that's fine you know you you can do that this but then i think you always need to go in and check the sort less citation and then pull up the article so you can read it yourself. So I do think it's a great tool to find articles because it makes them so easy to find and answer your research question. But then you need to go in and look at the article because a lot of times it could be miscited or misreferenced.
josh MacDonald: Yeah, yeah, definitely. one of I think even before AI existed, we've had Google Scholar. It's just scholar .google .com. It's a fantastic way to put in a general search like a Google search. and
josh MacDonald: And it gives us information that is like just article links. And usually you can read the abstract, but you may not have more than that. But it gives you lots of information about. that topic and you can look at article after article it doesn't give you a summary but I don't know how much I trust the summary of it.
josh MacDonald: I want to decide that for myself. So Google Scholar is a great avenue for finding out like hey what is the recovery likelihood of a proximal carpectomy what are the range of motions that i'm going to see that's oftentimes in an abstract of an an actual peer review article.
Miranda Materi: Yeah, and you know, AI can search so much. more quickly and all the data, but can you imagine us manually trying to find a lot of this information like we used to, right? We would pull out Rehab of the Hand and then we would, you know, email the librarian.
Miranda Materi: And so it's just so much more efficient now.
josh MacDonald: Yeah. Yeah.
Miranda Materi: You've saved so much time, I think. it's a little bit easier to utilize evidence -based, but I also think you have to be a little bit more Um.
Miranda Materi: willing to critically look at. what you're looking up.
josh MacDonald: Yeah. And we've there's a story that came out maybe a week ago um that was about the NHS, the National... ah I'm going to forget the name of the actually, but the the federal government's department that is in charge of of putting out research and publications on.
josh MacDonald: um health topics the and NHS they were, um, They put out a press release and it had referenced articles. And like we should, someone went and checked what those articles actually said. And they were not actual articles.
josh MacDonald: Someone had plugged in and said, give me article references that support this concept. And they they weren't actual articles that ChatGPT or whichever format they use just gave them references.
josh MacDonald: to articles that didn't exist. And so we have to be careful to make sure that I'm not Like it can only give me what I've asked for. I have to make sure I'm asking and giving a prompt that is going to give me. an actual article that exists and it's not a one -off you know we learn in statistics class you can find research that supports just about anything and we see this all the time about, you know, like on social media about studies that support this. Make sure it's an actual study and it's a good study.
Miranda Materi: Yeah. You know, I think that brings me... gave me an idea. I think it would be helpful as if there was like AI for the healthcare worker, right? Like what's HIPAA compliant? Where can you really find evidence based? How do you critically appraise and look at what AI is giving you?
Miranda Materi: you know, we're used to like critically appraising articles that you know come from google scholar but What about from AI?
josh MacDonald: Yeah.
Miranda Materi: How is that different?
josh MacDonald: And I think that's the world we live in now, where if I... If I see something, whether it's on social media or a news source or something, if I see it, Before I take it as factual or reality, I need to look up ah a corroborating source, go to some other source and search for it, even if I'm doing a general Google search. But if it's a research article. Go independently Google Scholar that research article and see if it pops up and see if it says what.
josh MacDonald: You're being told it says if you see one of our treatment guidelines for something and there's an article at the bottom. Go look up that article. Does it say what it said it did? Because we find sometimes when I read a protocol, I'm like, oh, let me see. what that article actually says.
josh MacDonald: And it's for something else. So we need to validate as. As information consumers, we need to be making sure that the things that we're being told are supported by other sources.
Miranda Materi: Yeah. Definitely.
josh MacDonald: Yeah. It's a crazy world with AI stuff going on and it's changing by the minute, but we can use it as a tool so long as it's not something that we just blindly rely on the first information it spits back at us for sure.
Miranda Materi: Yeah. And I think that's... brings up a good point. When you're a new therapist, it's very easy to, because you don't know, right? You don't know what you don't know. And like where we've been in this space for, you know, 20 years or whatever, it's easy for us to pick up mistakes, right?
Miranda Materi: So even like trying to find images is a big one, like AI will generate images, but there's so many mistakes on them. As a new grad, you might not know that, right?
josh MacDonald: Yeah.
Miranda Materi: You might not be like, you'd be like, oh yeah, that looks like trigger finger. And then it's in like the. you know a four pole or so you know something like that it's really in a different region so i think
josh MacDonald: Yeah, yeah
Miranda Materi: um it becomes even more important as a a new learner or a new grad to really be able to critically appraise and look at that stuff.
josh MacDonald: Yeah. And as AI gets better, I think those mistakes aren't necessarily disappearing they're just harder to find. And so we have to be more critical about it because right now you can tell, oh, that person has a third hand.
josh MacDonald: We've done that one trying to create. images for
Miranda Materi: yeah
josh MacDonald: for documents we're putting out like oh i'll ask ai to give me someone doing a know scaption plane shoulder flexion and there's the third arm on it so we could easily find that but as it gets better it may be more subtle things and we have to be more diligent about finding ah looking for those errors and mistakes in content, in images, in references, all that stuff.
josh MacDonald: Yeah. All right, so AI is a big confusing thing. We can use it as a tool, but we got to be careful with it. so Let us know what you think. If you have any thoughts or suggestions on it, you can reach us on social media at Hand Therapy Academy or on our email. info at hand therapy academy .com
